Analyses were performed separately within each dataset and population group, corrected for the subgroup-specific genomic inflation factor (λ), and the results were combined by meta-analysis using the inverse variance method. As described above, we then tested SNPs with a p value <1 × 10−4 in either population group or the full meta-analysis (n = 10,390) in a model adjusted for the DSM-IV criterion counts for cocaine, opioid, and alcohol dependence. We also tested 20,336 genotyped SNPs on the X chromosome and 226 on the Y chromosome for FTND association. Y chromosome SNPs were tested as binary variables in male subjects only and X chromosome SNPs were coded as homozygous in male subjects. A p value of 5.0 × 10−8 was the threshold for GWS in the GWAS; this applies a Bonferroni correction covering all independent haplotype blocks (regardless of the number of SNPs tested). Results were not adjusted for testing in two populations because we tested three distinct a priori hypotheses: SNPs are associated with FTND in AAs, SNPs are associated with FTND in EAs, and associations are evident with
